An XR virtual resource preloading method

Abstract

The application discloses an XR virtual resource preloading method, and relates to the field of extended reality, and comprises the following steps: setting a fixed numerical preloading space reference volume according to the computing power and memory parameters of a target XR device, and storing the preloading space reference volume to an XR device local configuration module; collecting the current space position, instantaneous travel speed and instantaneous orientation transformation rate of a user in a virtual scene in real time to obtain real-time motion pose parameters; the application sets a fixed preloading space reference volume based on the computing power and memory of an XR device, constructs a dynamically adjustable quadrangular pyramid preloading area along the orientation of the user, adjusts the shape in combination with real-time motion pose, maintains the constant total volume, and adopts incremental resource scheduling, so that the resource loading delay can be reduced, invalid memory occupation can be reduced, and the smoothness of XR virtual scene operation can be improved.

Description

Technical Field

[0001] This invention relates to the field of extended reality technology, specifically to a method for preloading XR virtual resources. Background Technology

[0002] XR applications require a large number of virtual resources to run, including 3D models, textures, materials, animations, and spatial audio. Due to the limitations of head-mounted displays or mobile devices in terms of storage bandwidth and computing power, if the required resources are loaded only after the user's gaze or interaction is triggered, screen stuttering, delayed object appearance, or even scene loss can easily occur. To reduce these problems, virtual resources that may be used are usually loaded into memory in advance before the interaction occurs, based on factors such as scene partitioning, preset paths, user history operations, or gaze direction.

[0003] Existing preloading methods mostly refer to the ideas of game streaming loading and network prefetching, and control the loading order through priority queues, cache management and memory budget to ensure that the required resources are basically ready when called.

[0004] However, in XR interaction, viewpoint movement, gesture operation and scene switching are highly spontaneous. Existing methods rely on fixed rules or historical statistics, which makes it difficult to accurately predict the next resource demand. This often leads to low preloading hit rate or excessive memory consumption, making it difficult to improve the preloading hit rate and control memory consumption while ensuring smooth interaction. Summary of the Invention

[0005] In view of the above-mentioned shortcomings of the existing technology, the present invention provides an XR virtual resource preloading method, which can effectively solve the problems of the existing technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions;

[0007] This invention discloses an XR virtual resource preloading method, comprising:

[0008] Based on the computing power and memory parameters of the target XR device, a fixed preload space reference volume is set and stored in the XR device's local configuration module. Real-time data collection is performed on the user's current spatial position, instantaneous travel speed, and instantaneous orientation change rate within the virtual scene to obtain real-time motion pose parameters. An initial quadrangular pyramid with a tilting posture is constructed along the user's current orientation. The triangular cross-section of this pyramid, including its axis, is determined, ensuring that the cross-section does not coincide with any lateral edge of the initial quadrangular pyramid, so that the user's current spatial position falls on the central axis of this cross-section. Using the preload space reference volume as a constraint, the height of the quadrangular pyramid is adjusted based on the instantaneous travel speed, and the base area of ​​the quadrangular pyramid is adjusted accordingly based on the instantaneous orientation change rate, ensuring that the volume of the quadrangular pyramid is always equal to the preload space reference volume. All resource units within the virtual scene are traversed, and all resource units falling within the space enclosed by the adjusted quadrangular pyramid are selected to generate a preload resource list. The corresponding virtual resources are loaded into the device memory according to the preload resource list, and the quadrangular pyramid enclosed space and the preload resource list are iterated synchronously with the real-time updates of the user's pose data.

[0009] Furthermore, a fixed preload space is set to remember the volume during the phase, and the virtual dimensions of the virtual scene are acquired simultaneously. Let the length, width, and height be x, y, and z, respectively. The voxel size a³ is set simultaneously, ensuring that x, y, and z are all divisible by a. The virtual scene is then divided using the voxel size. For each voxel, obtain the memory required for loading each voxel, and mark the memory required for loading each voxel on the corresponding voxel.

[0010] After setting a fixed preload space reference volume, the initial square pyramid shape and the number of voxels are determined. The initial square pyramid is determined to be a pyramid structure of approximately square pyramid shape composed of several voxels.

[0011] ;

[0012] In the formula: n represents the number of voxel stacking layers in the initial square pyramid; V represents the set preload space reference volume;

[0013] When the above equation holds true, the initial square pyramid contains the following number of voxels: If the length, width, and height of the initial quadrangular pyramid are all n×a, and the vertex of the initial quadrangular pyramid is located on the perpendicular line from the center point of the base of the initial quadrangular pyramid.

[0014] Furthermore, the preloaded space reference volume V is determined through the following steps:

[0015] Based on the memory required for loading each voxel, sort the voxels to obtain the voxel queue.

[0016] Extract the first 1 / 4 to 1 / 5 of the voxel queue, simultaneously identify the position of each extracted voxel in the virtual scene, and construct the minimum bounding box of all extracted voxels based on the identified position, using the volume of this minimum bounding box as V.

[0017] Furthermore, the acquisition frequency of the user's current spatial position, instantaneous travel speed, and instantaneous orientation change rate in the virtual scene is consistent with the acquisition frequency of motion pose parameters. The initial frequency is set to 0.1s / time, and is dynamically adjusted within the range of ±0.05s / time according to the instantaneous travel speed.

[0018] Furthermore, after the initial square pyramid is constructed, it is adjusted to a tilted posture along the user's current orientation:

[0019] The real-time motion pose parameters are based on the user's movement path established by continuously acquiring the user's position coordinates in the virtual scene, and the user movement path contains continuously acquired user position coordinate nodes.

[0020] Get the path segment formed by the latest two nodes on the user's movement path. and based on Determine the direction of movement;

[0021] Tilting the initial square pyramid so that the axis passing through the vertex of the initial square pyramid is aligned with... Overlapping, and based on The determined direction of movement is directly opposite the center point of the base of the initial square pyramid after it has tipped over;

[0022] Based on the intersection of the base and vertex of a square pyramid, the axis passing through the vertex is denoted as... The captured data will be determined based on the direction of movement. The two ends are set as the start and end points. The initial position of the user in the quadrangular pyramid in the virtual scene is set to 3 / 4 of the distance from the end point and 1 / 4 of the distance from the start point, and is dynamically adjusted based on the user's current turnaround rate in the virtual scene.

[0023] Furthermore, when adjusting the height of the pyramid based on the instantaneous travel speed, the base area of ​​the pyramid is determined by combining the constraint that the volume of the pyramid is always equal to the preloaded space reference volume.

[0024] The adjustment of the pyramid height based on instantaneous travel speed follows:

[0025] ;

[0026] In the formula: The forward extension height of the square pyramid; Instantaneous speed; This refers to the average loading response latency per resource unit of an XR device. The maximum travel acceleration set for users in a virtual environment; This is the adjusted height of the square pyramid; This is the initial reference height of the square pyramid; This is the adjusted base area of ​​the square pyramid; Preload space reference volume;

[0027] Among them, the faster the instantaneous speed, the greater the forward extension height of the square pyramid, the higher the total height, and the smaller the corresponding base area.

[0028] Furthermore, based on the instantaneous orientation change rate corresponding to the adjustment of the base area of ​​the pyramid, and combined with the constraint that the volume of the pyramid is always equal to the preloaded spatial reference volume, the height of the pyramid is determined:

[0029] The adjustment of the base area of ​​the quadrangular pyramid based on the instantaneous orientation change rate follows the following:

[0030] ;

[0031] In the formula: The lateral expansion angle of the base of the square pyramid; The instantaneous orientation change rate; This refers to the average loading response latency per resource unit of an XR device. The maximum user viewpoint acceleration set for the virtual scene; This is the adjusted base area of ​​the square pyramid; The base area of ​​the initial square pyramid; The horizontal field of view angle is the reference angle for XR equipment; This is the adjusted height of the square pyramid; Preload space reference volume;

[0032] Among them, the faster the instantaneous orientation change rate, the larger the lateral expansion angle of the base of the square pyramid, the larger the total base area, and the smaller the corresponding height.

[0033] Furthermore, during the dynamic adjustment process, the square pyramid selectively chooses either a height adjustment logic based on instantaneous travel speed or a base area adjustment logic based on instantaneous orientation change rate to perform a shape update. The selection logic is as follows:

[0034] The target cone height is calculated by combining the current instantaneous travel speed. The absolute value of the difference between the current pyramid height and the target cone height is calculated, and the ratio of this difference to the current pyramid height is used to obtain the height deviation rate.

[0035] The target base area is calculated by combining the current instantaneous orientation change rate. The absolute value of the difference between the current base area of ​​the pyramid and the target base area is calculated, and the ratio of this difference to the current base area of ​​the pyramid is used to obtain the base area deviation rate.

[0036] Obtain the current user's motion return rate, use the return rate as the lateral weight coefficient, and use the difference between 1 and the return rate as the axial weight coefficient. Weight the bottom area deviation rate and the height deviation rate respectively to obtain the lateral weighted deviation value and the axial weighted deviation value.

[0037] Compare the values ​​of the two sets of weighted deviation values, and select the dimension corresponding to the larger weighted deviation value for this morphological update;

[0038] When the absolute value of the difference between the two sets of weighted deviation values ​​is less than the preset hysteresis threshold, the adjustment logic used in the previous frame is maintained as the calculation rule for this morphological update.

[0039] Furthermore, when filtering all resource units that fall within the space enclosed by the adjusted quadrangular pyramid, the voxels obtained by voxel segmentation in the virtual scene are used as the smallest unit. All complete voxels contained in the quadrangular pyramid are identified, and the resource units corresponding to each voxel are used as preloading targets to generate a preloaded resource list.

[0040] Furthermore, the steps of synchronously iterating the quadrangular pyramid bounded space and the preloaded resource list with the real-time update of user pose data include:

[0041] After each adjustment of the quadrangular pyramid shape and generation of the adjusted preloaded resource list, the preloaded resource list before adjustment and the preloaded resource list after adjustment are compared. The overlapping resource units in the two lists are retained and their loading status is maintained.

[0042] All resource units in the pre-loaded resource list that do not belong to the overlapping part will be removed from the loading process;

[0043] Perform loading operations on newly added resource units that do not belong to the overlapping part in the adjusted preloaded resource list;

[0044] After all resource units in the preloaded resource list have been loaded after the adjustment, the process will proceed to the next step of determining overlapping resource units after the adjustment of the quadrangular pyramid shape.

[0045] Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects:

[0046] This invention sets a fixed preload space baseline volume based on the computing power and memory parameters of XR devices. It performs voxel segmentation of the virtual scene and marks the memory usage of each voxel. A quadrangular pyramid with a tilted posture is constructed along the user's current orientation as the preload enclosing space. The height and base area of ​​the pyramid are dynamically adjusted based on the real-time collected user position, movement speed, and orientation change rate to maintain a constant total volume of preload space to adapt to the device's memory capacity limit. A weighted matching adjustment dimension is introduced based on the user's movement turnaround rate. With a preset hysteresis threshold, the computational overhead caused by frequent form switching is reduced. An incremental loading and unloading mechanism is used to retain the state of overlapping resources. Under limited memory conditions, virtual resources can be scheduled according to the user's movement trend, reducing resource loading response latency, reducing the memory usage of invalid preload, and improving the smoothness of virtual scene operation and the rationality and hit rate of resource pre-scheduling loading. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0048] Figure 1 This is a flowchart illustrating a method for preloading XR virtual resources.

[0049] Figure 2 This is a schematic diagram illustrating the cross-section of a quadrangular pyramid, its central axis, and the user's direction of travel in the virtual scene of this invention.

[0050] Figure 3 This is a two-dimensional illustration of the distribution of voxels used in the initial construction of the quadrangular pyramid in this invention from the side, top, and bottom perspectives.

[0051] Figure 4 This is a two-dimensional view of the side, top, and bottom of the quadrangular pyramid obtained after the initial quadrangular pyramid is tilted in this invention.

[0052] Figure 5 This is a comparison curve of the real-time computing power and memory consumption of the XR virtual resource preloading method in this invention with that of existing preloading methods. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0054] The present invention will be further described below with reference to embodiments.

[0055] Example:

[0056] This embodiment provides a method for preloading XR virtual resources, such as... Figure 1 As shown, it includes:

[0057] Based on the computing power and memory parameters of the target XR device, a fixed preload space baseline volume is set and stored in the XR device's local configuration module;

[0058] Among them, the computing power parameter indirectly affects the final value of the base volume through the voxel granularity. The higher the computing power, the smaller the voxel size that can be supported, the more effective voxels are contained in the same bounding box volume, and the higher the accuracy of the base volume setting. At the same time, when the computing power is sufficient, the voxel truncation ratio can be appropriately increased to increase the base volume within the memory allowance and improve the preload coverage.

[0059] During the preloading phase, a fixed value is set for the space to remember the volume. Simultaneously, the virtual dimensions of the virtual scene are acquired. Let the length, width, and height be x, y, and z, respectively. Simultaneously, the voxel size a³ is set, ensuring that x, y, and z are all divisible by a (voxel size). The value needs to be determined by combining the computing power of the XR device and the precision of scene resources. The stronger the computing power and the finer the granularity of scene resources, the better. The smaller the value, the better; the typical value range is 0.5 to 2m. When setting, first take the greatest common divisor of the scene's length, width, and height as the initial selection value, and then check whether the traversal calculation time corresponding to the total number of voxels under this size is less than one-third of the budgeted time for a single frame. If it exceeds this, increase the value. The value of is determined until the computing power constraint is met, while ensuring that all dimensions of the scenario can be utilized. (Divisible by integer), the virtual scene is divided by voxel size into For each voxel, obtain the memory required for loading each voxel, and mark the memory required for loading each voxel on the corresponding voxel.

[0060] The memory required for voxel loading is calculated based on all virtual resources at the highest precision level within that voxel, including the total memory usage of 3D models, texture maps, materials, skeletal animations, and spatial audio. The actual running memory after resource decompression is included in the calculation to ensure that the memory budget is consistent with the actual running state.

[0061] After setting a fixed preload space reference volume, the initial square pyramid shape and the number of voxels are determined. The initial square pyramid is determined to be a pyramid structure of approximately square pyramid shape composed of several voxels.

[0062] ;

[0063] In the formula: n represents the number of voxel stacking layers in the initial square pyramid; V represents the set preload space reference volume;

[0064] When the above equation holds true, the initial square pyramid contains the following number of voxels: If the length, width and height of the initial quadrangular pyramid are all n×a, and the vertex of the initial quadrangular pyramid is located on the perpendicular line from the center point of the base of the initial quadrangular pyramid;

[0065] The dynamic adjustment process uniformly adopts the geometric square pyramid volume formula for calculation, with the volume benchmark being the preloaded space benchmark volume. ;

[0066] Reference height of the initial square pyramid Pick ,in The number of voxel stacking layers. Let the side length of the voxel be the value; then, according to the formula for the volume of a square pyramid... The initial reference base area is obtained by reverse calculation. ;

[0067] The voxel stacking pattern is used for discretization adaptation in the initial stage. Subsequent dynamic adjustments are all based on continuous geometric parameters to ensure the consistency of volume constraints.

[0068] This formula solves for the number of voxel stacking layers of a quadrangular pyramid under the constraint of a fixed preloaded spatial reference volume by accumulating and comparing voxel stacks. It simulates the actual volume occupied by the pyramid-shaped voxel stack based on the sum of squares, discretizes the continuous spatial volume into a combination of standardized voxel units, avoids the floating-point operation overhead caused by directly solving the geometric parameters, and allows the initial quadrangular pyramid to adapt to the virtual scene after voxel segmentation. This ensures that the memory information marked by voxels can be directly reused in subsequent resource screening, and achieves a unified measurement of spatial boundaries and resource memory usage.

[0069] The preloaded space reference volume V is determined through the following steps:

[0070] Based on the memory required for loading each voxel, sort the voxels (from high to low) to obtain the voxel queue;

[0071] Extract the first 1 / 4 to 1 / 5 of the voxel queue, simultaneously identify the position of each extracted voxel in the virtual scene, and construct the minimum bounding box of all extracted voxels based on the identified position, using the volume of this minimum bounding box as V.

[0072] The truncation ratio of the voxel truncation queue is set according to the performance of the XR device and the user's XR experience requirements. The better the performance of the XR device and the higher the user's XR experience requirements, the larger the truncation ratio will be, and vice versa.

[0073] In actual setup, first read the maximum available memory allocated by the XR device to the virtual scene preloading function. Statistical analysis of the average memory usage of all elements ,by Obtain the maximum number of voxels that can be carried. ;

[0074] Then calculate the total number of elements. of and The corresponding value will Compare with this interval: If Greater than Then the truncated proportion is taken as ;like Less than Then the truncated proportion is taken as If it is within the range, then proceed as follows: The ratio of the nearest match to The ratio between these values ​​ensures that the total memory of resources corresponding to the baseline volume does not exceed the device's available limit.

[0075] Real-time data collection of the user's current spatial position, instantaneous travel speed, and instantaneous orientation change rate within the virtual scene to obtain real-time motion pose parameters;

[0076] The frequency of collecting the user's current spatial position, instantaneous travel speed, and instantaneous orientation change rate in the virtual scene is consistent with the frequency of acquiring motion pose parameters. The initial frequency is set to 0.1s / time, and is dynamically adjusted within the range of ±0.05s / time according to the instantaneous travel speed.

[0077] Among these, the frequency dynamic adjustment follows the rule that the greater the instantaneous rate of change of travel speed, the higher the dynamic adjustment amount:

[0078] ;

[0079] In the formula: The adjusted frequency; For dynamic adjustment; This represents the current instantaneous speed of travel. The previous instantaneous speed, when If the timeout is not within the range of [0.05s / timeout, 0.15s / timeout], take the current timeout value. The nearest value in the range of possible values;

[0080] The above formula uses 0.1 seconds as the basic period for pose acquisition. It dynamically corrects the acquisition frequency by combining the changes in instantaneous movement speed between two consecutive moments. It converts the relative change in speed into a frequency adjustment amount and limits the acquisition period to the range of 0.05-0.15 seconds for boundary clamping. When the user's movement is stable, it maintains a low sampling update overhead. When the movement speed changes drastically, it increases the sampling sensitivity. This avoids high-frequency sampling from continuously consuming the device's computing power and can capture pose changes in a timely manner in scenarios where the user's actions change abruptly, preventing lag deviations in the preload space.

[0081] Construct an initial quadrangular pyramid with a tilting posture along the user's current orientation, determine the triangular cross-section of the pyramid containing the pyramid axis, and ensure that the cross-section does not coincide with any of the lateral edges of the initial quadrangular pyramid, so that the user's current spatial position falls on the central axis of the cross-section;

[0082] At the implementation level, the user's current orientation refers to the user's line of sight, that is, the direction directly in front of the view output by the head-mounted display; the movement direction is the direction of movement corresponding to the change in the user's position, which is an auxiliary adjustment parameter. When the user has an effective movement speed and the angle between the movement direction and the line of sight is less than 30°, the cone axis coincides with the movement direction; when the angle is greater than or equal to 30°, or when the user is stationary and has no effective movement direction, the cone axis is uniformly based on the line of sight to ensure that the preloaded space prioritizes covering the user's visible range;

[0083] This triangular cross-section serves as the symmetrical reference plane for adjusting the shape of the quadrangular pyramid. When expanding the base area based on the orientation transformation rate, the expansion is symmetrically directed to both sides from this cross-section as the center, ensuring the uniformity of the base area expansion. The cross-section is required not to coincide with the side edges. This is to limit the user's position relative to the quadrangular pyramid from falling on the intersection and edge, preventing the user from having no data loaded in a certain direction in the quadrangular pyramid space in the virtual field. This avoids expansion calculation errors caused by the overlap of a single boundary with the side edges when expanding the base area, ensuring the accuracy of the lateral angle adjustment.

[0084] After the initial square pyramid is constructed, it is adjusted to a tilted position according to the user's current orientation:

[0085] The real-time motion pose parameters are based on the user's movement path established by continuously acquiring the user's position coordinates in the virtual scene. The user's movement path contains continuously acquired user position coordinate nodes.

[0086] Get the path segment formed by the latest two nodes on the user's movement path. and based on Determine the direction of movement;

[0087] When the instantaneous travel speed is detected as If the data collection continues for more than two consecutive collection cycles, the user is determined to be stationary. At this point, the pyramid's axis coincides with the user's current line of sight, and the pyramid's height and base area are restored to their initial baseline values. and At the same time, the user's position on the cone axis is adjusted to half the height of the cone from the bottom surface to ensure uniform resource coverage around the perimeter when stationary.

[0088] Tilting the initial square pyramid so that the axis passing through the vertex of the initial square pyramid is aligned with... Overlapping, and based on The determined direction of movement is directly opposite the center point of the base of the initial square pyramid after it has tipped over;

[0089] Based on the intersection of the base and vertex of a square pyramid, the axis passing through the vertex is denoted as... The captured data will be determined based on the direction of movement. The two ends are set as the start and end points (Note: the end point along the cone axis in the forward direction of preloading is the end point, corresponding to the center point of the base of the pyramid; the end point of the cone axis facing the user's rear is the start point, corresponding to the vertex of the pyramid. The user's facing direction and movement direction both point towards the end point, ensuring that the preloaded space extends further towards the user's movement and field of vision). The user's initial position in the pyramid within the virtual scene is set at 3 / 4 of the distance from the end point and 1 / 4 of the distance from the start point, and dynamically adjusted based on the user's current foldback rate in the virtual scene.

[0090] The system counts several sub-paths obtained by segmenting the user's movement path based on the user's location coordinate nodes, determines the attribute of each sub-path as forward or back, and uses the ratio of the number of sub-paths with the attribute of back to the total number of sub-paths as the back rate.

[0091] In this process, the vector from the start point to the end point of each sub-path is taken as the direction of travel of that sub-path. When determining a turnaround, the angle between the current sub-path and the direction of travel of the previous sub-path is calculated. If the angle is greater than a preset angle, the sub-path is determined to be a turnaround path.

[0092] The higher the turnaround rate, the more the initial position moves towards the endpoint; the lower the turnaround rate, the more the initial position moves towards the starting point. The minimum constraint between the initial position and the endpoint is set to 1 / 2, and the minimum constraint between the initial position and the starting point is set to 1 / 5.

[0093] The ratio of the user's location to the destination is calculated using a linear mapping method, and the turnaround rate is... When the position is 3 / 4 from the end point and 1 / 4 from the start point, and the turnaround rate is 1, the position is 1 / 2 from the end point and 1 / 2 from the start point. The entire course meets the constraint requirement that the distance from the start point is not less than 1 / 5 and the distance from the end point is not less than 1 / 2. The adjustment process changes continuously and linearly with the turnaround rate.

[0094] During the dynamic adjustment of the quadrangular pyramid shape, the user's position on the pyramid axis always maintains the value corresponding to the current foldback rate, that is, the proportion from the bottom end point remains unchanged.

[0095] Using the preloaded space reference volume as a constraint, the height of the quadrangular pyramid is adjusted based on the instantaneous travel speed, and the base area of ​​the quadrangular pyramid is adjusted accordingly based on the instantaneous orientation change rate, and the volume of the quadrangular pyramid is always equal to the preloaded space reference volume.

[0096] When adjusting the height of the pyramid based on the instantaneous travel speed, the base area of ​​the pyramid is determined by combining the constraint that the volume of the pyramid is always equal to the preloaded space reference volume.

[0097] The height of the pyramid is adjusted based on instantaneous travel speed according to:

[0098] ;

[0099] In the formula: The forward extension height of the square pyramid; Instantaneous speed; This refers to the average loading response latency per resource unit of an XR device. The maximum travel acceleration set for users in a virtual environment; This is the adjusted height of the square pyramid; This is the initial reference height of the square pyramid; This is the adjusted base area of ​​the square pyramid; Preload space reference volume;

[0100] Average load response time per resource unit This can be obtained through local pre-testing on the device. Select a resource unit of typical size within the scene and load it continuously 10 times, then take the average loading time as the result. The typical value range is to Maximum user acceleration With maximum angular acceleration The interaction specifications for virtual scenes are set. Typical value to , Typical value to The adjustment can be made up or down according to the intensity of interaction in the scene;

[0101] The above formulas combine instantaneous travel speed, device resource loading latency, and maximum travel acceleration in the scene to calculate the height increment of the pyramid extending forward. Under the constraint of a constant preloaded baseline volume, the corresponding base area is obtained by back-calculating the pyramid height. The predictable distance that the user can reach forward is mapped to the axial expansion of the pyramid. The faster the user travels, the greater the forward extension of the pyramid. Simultaneously, the base area is compressed, and virtual resources in areas further away in the direction of travel are preloaded first to match the user's scene access expectations when moving forward. At the same time, the volume constraint controls the total amount of preloaded resources to not exceed the device's set upper limit.

[0102] Among them, the faster the instantaneous speed, the greater the forward extension height of the square pyramid, the higher the total height, and the smaller the corresponding base area;

[0103] When adjusting the base area of ​​the pyramid based on the instantaneous orientation change rate, the height of the pyramid is determined by combining the constraint that the volume of the pyramid is always equal to the preloaded space reference volume:

[0104] When the device has a roll angle around the line of sight axis, the entire pyramid rotates synchronously with the roll angle, keeping the sides of the square base aligned with the horizontal and vertical directions of the user's current field of vision, ensuring that the lateral expansion direction is consistent with the direction the user turns their head, and avoiding the preload area offset caused by the roll.

[0105] The adjustment of the base area of ​​the square pyramid based on the instantaneous orientation change rate follows the following principle:

[0106] ;

[0107] The adjustment of the base area uses a two-step calculation logic:

[0108] First, assume that the cone height remains at the initial reference value. The theoretical target base area is obtained by combining the lateral angle increment with the unchanged lateral angle increment.

[0109] With constant volume as a constraint, the actual adjusted cone height is obtained by back-calculating the theoretical target base area. The final shape meets the requirement of constant volume. The angular expansion is slightly reduced as the cone height decreases. Under the premise of ensuring controllable memory, it approximately matches the coverage requirement of the viewpoint rotation.

[0110] In the formula: The lateral expansion angle of the base of the square pyramid; The instantaneous heading change rate is the combined angular velocity of horizontal yaw and vertical pitch.

[0111] Instantaneous orientation change rate The resultant angular velocity is calculated as the square root of the sum of the squares of the horizontal yaw rate and the vertical pitch rate, with units of 1000 rpm. The calculation of the lateral expansion angle of the base area only uses the horizontal yaw rate component. The vertical pitch rate is used to synchronously adjust the vertical angle of the pyramid. Due to the limited field of view in the vertical direction, the adjustment range is based on the horizontal direction. The calculation is multiplied to ensure that the resource coverage of the upper and lower field of view matches the head-turning action;

[0112] This refers to the average loading response latency per resource unit of an XR device. The maximum user viewpoint acceleration set for the virtual scene; This is the adjusted base area of ​​the square pyramid; The base area of ​​the initial square pyramid; The horizontal field of view angle is the reference angle for XR equipment; This is the adjusted height of the square pyramid; Preload space reference volume;

[0113] The above formula sets combine the angular velocity and angular acceleration corresponding to the user's viewpoint rotation with the device loading delay to obtain the lateral angle increment of the viewpoint. Combined with the tangent ratio of the reference viewpoint angle, the base area is converted. Then, the matching cone height is solved by the fixed preload volume. When the user turns his head and tilts his viewpoint, the lateral coverage of the base of the pyramid expands accordingly, and the cone height is reduced accordingly. It prioritizes covering the lateral scene area that the user is about to look at. Under the premise of keeping the total preload volume unchanged, it adapts to the resource prediction requirements when the viewpoint rotates rapidly without increasing the device memory load.

[0114] Among them, the faster the instantaneous orientation change rate, the larger the lateral expansion angle of the base of the square pyramid, the larger the total base area, and the smaller the corresponding height;

[0115] The bottom area adjustment formula is derived based on the regular square pyramid structure. The bottom surface always maintains a square shape. The horizontal and vertical directions are calculated using a uniform lateral angle. The resultant angular velocity of the instantaneous orientation change rate is uniformly converted into the lateral angle increment. By simplifying geometric calculations, the computing power of the equipment is reduced, and the operating performance constraints of the XR end are adapted.

[0116] During the dynamic adjustment process of the square pyramid, either the height adjustment logic based on the instantaneous travel speed or the base area adjustment logic based on the instantaneous orientation change rate is selected to perform a shape update. The selected logic is as follows:

[0117] The target cone height is calculated by combining the current instantaneous travel speed. The absolute value of the difference between the current pyramid height and the target cone height is calculated, and the ratio of this difference to the current pyramid height is used to obtain the height deviation rate.

[0118] The target base area is calculated by combining the current instantaneous orientation change rate. The absolute value of the difference between the current base area of ​​the pyramid and the target base area is calculated, and the ratio of this difference to the current base area of ​​the pyramid is used to obtain the base area deviation rate.

[0119] Obtain the current user's motion return rate, use the return rate as the lateral weight coefficient, and use the difference between 1 and the return rate as the axial weight coefficient. Weight the bottom area deviation rate and the height deviation rate respectively to obtain the lateral weighted deviation value and the axial weighted deviation value.

[0120] Compare the values ​​of the two sets of weighted deviation values, and select the dimension corresponding to the larger weighted deviation value for this morphological update;

[0121] Each time, only the dimension with the larger deviation weighting value is adjusted, while the deviation of the other dimension is retained and included in the next deviation rate calculation. As the user's motion state changes, when the weighted deviation value of the previously unadjusted dimension surpasses the previous one, the corresponding dimension is adjusted. The form adaptation of the two dimensions is completed through successive approximations, avoiding the computational overhead and loading jitter caused by adjusting both dimensions simultaneously within a single frame.

[0122] When the absolute value of the difference between the two sets of weighted deviation values ​​is less than the preset hysteresis threshold, the adjustment logic used in the previous frame is maintained as the calculation rule for this morphological update.

[0123] The first frame of the XR device when it is first launched uses the user's current line of sight as the direction of the cone axis by default. The four-sided pyramid uses the initial reference height and reference base area. The user's position on the cone axis is taken as the default distance of three-quarters of the distance to the end point.

[0124] The initial morphological update prioritizes the height adjustment logic based on instantaneous travel speed. After accumulating more than two frames of pose data, the adjustment dimension is selected according to the deviation weighting rule.

[0125] The preset hysteresis threshold is initially set to a dimensionless constant between 0.04 and 0.09, which is pre-configured by the XR device according to the interaction stability requirements.

[0126] Traverse all resource units in the virtual scene, filter all resource units that fall within the space enclosed by the adjusted quadrangular pyramid, and generate a preloaded resource list.

[0127] After determining the parameters of the geometric square pyramid, first establish the spatial equation of the pyramid, using the user's current position as the reference point, along the pyramid axis according to the voxel side length. The step size is divided into layers, each layer corresponds to a square cross section perpendicular to the cone axis. For each layer cross section, the side length of the square is calculated, and the number of whole rows and columns of voxels contained within the side length range is counted. The set of all complete voxels is accumulated layer by layer, which is the voxel range corresponding to the space enclosed by the square pyramid. This is used to match the resource unit to generate a preload list.

[0128] When filtering all resource units that fall within the space enclosed by the adjusted quadrangular pyramid, the voxels obtained by voxel size segmentation in the virtual scene are used as the smallest unit. All complete voxels contained in the quadrangular pyramid are identified, and the resource units corresponding to each voxel are used as preloading targets to generate a preloaded resource list.

[0129] When the space enclosed by the quadrangular pyramid exceeds the boundary of the virtual scene, the scene boundary is used as the cut-off surface, and only the enclosed space inside the scene is retained. When filtering resources, only voxels within the scene range are traversed. If there is no corresponding voxel in the out-of-bounds area, it will not be included in the preloading range, ensuring that the preloading list only contains valid resources existing in the scene.

[0130] The resource unit division and voxel segmentation of the virtual scene correspond one-to-one, with a single resource unit corresponding to all virtual resources within a single voxel;

[0131] If there are independent large resources spanning multiple voxels, the resource bounding box is used as the unit for determination. When the central voxel of the resource bounding box falls into the space enclosed by the square pyramid, the resource is included in the preload list to avoid repeated splitting and loading.

[0132] When determining that a voxel is completely within the space enclosed by a square pyramid, the coordinates of the eight vertices of the voxel are verified one by one. If all eight vertices are inside the square pyramid or on its boundary surface, the voxel is determined to be a complete voxel and included in the preloaded resource list. If any vertex is outside the square pyramid, it is not included in the current preload range, ensuring that the spatial boundary of the preloaded resources is accurate and controllable.

[0133] The corresponding virtual resources are loaded into the device memory according to the preloaded resource list, and the quadrangular pyramid enclosed space and the preloaded resource list are synchronously iterated as the user pose data is updated in real time.

[0134] The steps for synchronously iterating the bounded space of the quadrangular pyramid and the preloaded resource list in real time with the user pose data include:

[0135] After each adjustment of the quadrangular pyramid shape and generation of the adjusted preloaded resource list, the preloaded resource list before adjustment and the preloaded resource list after adjustment are compared. The overlapping resource units in the two lists are retained and their loading status is maintained.

[0136] All resource units in the pre-loaded resource list that do not belong to the overlapping part will be removed from the loading process;

[0137] Perform loading operations on newly added resource units that do not belong to the overlapping part in the adjusted preloaded resource list;

[0138] After all resource units in the preloaded resource list have been loaded after the adjustment, the process will proceed to the next step of determining overlapping resource units after the adjustment of the pyramid shape.

[0139] If the previous round of resource loading has not been completed when a new round of pose update arrives, the unfinished loading task will be terminated immediately. Based on the preload list corresponding to the latest pose, overlapping resources will be re-compared and incremental loading and unloading will be performed. Resources will always be scheduled based on the latest pose data to avoid preload delays caused by loading queue accumulation.

[0140] In this embodiment, a fixed preload space baseline volume is set and stored in the local configuration module based on the computing power and memory parameters of the target XR device. The overall size of the virtual scene is obtained synchronously, and the scene is divided into several regular voxel sizes by setting a uniform voxel size. Each voxel is marked with the memory usage required for resource loading. The baseline volume can be determined by combining the distribution of voxels with high memory usage in the scene. The voxels with the highest memory usage are extracted to construct the minimum bounding box. The bounding box volume is used as the baseline value. The stronger the device performance and the higher the experience requirements, the larger the extraction ratio. This ensures the loading effect while adapting to the upper limit of device hardware and avoiding memory overload.

[0141] After initialization, the device collects the user's motion data in the virtual scene in real time, acquiring the user's current spatial position, instantaneous travel speed, and instantaneous orientation change rate at an initial frequency of 0.1 seconds to form real-time motion pose parameters. The acquisition frequency will be dynamically adjusted within ±0.05 seconds according to the change in instantaneous travel speed. The more drastic the speed change, the more significant the increase in acquisition frequency, ensuring the response accuracy of pose data and avoiding unnecessary computing power consumption in low-speed states.

[0142] Based on the current pose, an initial preloaded bounding space is constructed. A tilted quadrangular pyramid is built along the user's current orientation, so that the pyramid axis coincides with the user's movement direction and the user's current position falls on the pyramid axis. The initial position is at three-quarters of the pyramid length away from the base by default. At the same time, the back-turn rate of the user's historical movement path is calculated. The higher the back-turn rate, the further the position is offset from the base of the pyramid, ensuring that the resources in the user's field of vision can still be effectively covered when the user turns around. The initial quadrangular pyramid is formed by stacking voxels layer by layer to form an approximate pyramid shape, and the total volume strictly matches the preset baseline volume.

[0143] The shape of the pyramid is then dynamically adjusted based on the real-time pose, while maintaining a constant overall volume. If the user's movement speed increases, the pyramid height is lengthened along the direction of movement, while the base area is correspondingly reduced to ensure sufficient pre-loading distance for resources in the forward field of view. If the user's orientation change rate increases, the lateral angle of the pyramid's base area is expanded, while the pyramid height is correspondingly shortened to cover a wider lateral field of view. Before each shape update, the target deviation rate of the height and base area is calculated separately. After weighting by the return rate, the dimension with the larger deviation is selected for adjustment. If the difference between the two types of deviations is too small, the previous adjustment logic is used to avoid loading jitter caused by frequent dimension switching.

[0144] Once the quadrangular pyramid shape is determined, all resource units within the virtual scene are traversed. Using voxels as the smallest unit, all resources falling within the bounding space are selected, a preloaded resource list is generated, and memory loading is completed. After each round of pose data update, the quadrangular pyramid shape and resource list are iterated synchronously. The overlapping parts of resources in the new and old lists are compared, the loading status of overlapping resources is retained, resources that are no longer needed are unloaded, and newly added target resources are loaded. This achieves a smooth transition of visual resources within a limited memory space, effectively reducing loading stutters during scene switching and improving the overall smoothness of XR interaction.

[0145] See Figure 2 As shown in the diagram, the shaded area represents a cross-section of a square pyramid, the dashed line represents the central axis, and the arrows further indicate the user's direction of travel. This diagram aims to convey:

[0146] The user is positioned on the central axis of a quadrangular pyramid within the virtual scene (the central axis segment intercepted by the base and vertices of the quadrangular pyramid).

[0147] The quadrangular pyramid contains spatial resource units and XR device pre-loaded portions within the virtual scene.

[0148] See Figure 3 and Figure 4 As shown in the figure, the initial quadrangular pyramid structure is presented in two dimensions. It is actually an approximate quadrangular pyramid structure in which multiple voxels are stacked in a pyramid shape. By extending the bottom corner of the pyramid and the center point of the top voxel's top surface upwards, it is fitted into a quadrangular pyramid and further presented in a tilted posture. This shape is used to constrain the loading area of ​​preloaded resources. Subsequent adjustments also maintain the quadrangular pyramid shape, thereby saving computing power and memory consumption for preloaded resources and improving the rationality and smoothness of preloaded virtual resources.

[0149] See Figure 5 As shown, the graph has two vertical axes, with the left side representing the percentage of computing power consumption and the right side representing memory consumption (MB). The vertical axis represents the system running time (s). The legend contains four curves, which are the computing power consumption of the method of the present invention, the computing power consumption of the existing preloading method, the memory consumption of the method of the present invention, and the memory consumption of the existing preloading method.

[0150] As can be seen from the curve changes, the computing power and memory curves of the existing preloading method fluctuate drastically throughout the entire operation cycle, with frequent peak surges. The existing solution has insufficient accuracy in predicting resource demand, resulting in a large number of invalid preloadings. Repeated recalculation of resources and loading and unloading are performed, leading to computing power oscillations and drastic fluctuations in memory load.

[0151] In this embodiment, the method is based on a fixed preloaded space reference volume constraint. Only a single dimension is selected to perform morphological updates in a single frame. The method relies on the hysteresis threshold to suppress frequent parameter switching. Combined with incremental resource scheduling, the computing power curve fluctuates smoothly and the peak value is significantly lower. Memory consumption remains in a stable range without sudden spikes.

[0152] Comparative studies have shown that the present invention can smoothly control computing power and memory consumption, reduce the instantaneous load on XR edge hardware, alleviate device performance bottlenecks, reduce lag and heat generation, and verify the superiority of the resource scheduling of the present invention.

[0153] In the specific application scenario of the method in this embodiment, a square pyramid is selected as the basic shape of the preloaded enclosing space, which has multiple targeted technical advantages:

[0154] Firstly, the square base shape of the pyramid is highly compatible with the rectangular field of view distribution of the XR device. The direction of the pyramid axis extension corresponds to the user's walking and line of sight, and the lateral angle of the base corresponds to the horizontal and vertical field of view range. This can maximize the fit with the user's actual resource access area and effectively reduce the resource preloading overhead of the edge invalid space.

[0155] Secondly, the shape of the square pyramid can be defined using only two core parameters: the pyramid height and the base area. The volume calculation formula is simple, and the two parameters can be quickly calculated based on the constraint of a constant preloaded space reference volume during dynamic adjustment. The geometric calculation complexity is low, which can adapt to the limited computing power of XR edge devices and ensure the real-time performance of shape updates and resource selection within a single frame.

[0156] Third, the layered cross section of the quadrangular pyramid is a regular square, which is naturally aligned with the virtual scene mesh after voxel segmentation. The boundary calculation logic for voxel falling into the grid is simple, which can further reduce the computation time of spatial traversal and resource selection.

[0157] In practical applications, the core concept of this solution is not limited to the square pyramid shape. Depending on the field of view characteristics and computing power conditions of the specific scenario, other regular pyramid structures such as triangular pyramids, pentagonal pyramids, and hexagonal pyramids can be used as the basic shape of the preloaded enclosing space. Correspondingly, the voxel units for spatial segmentation can be replaced with regular segmentation units such as regular tetrahedrons that are suitable for the corresponding pyramid shape. For narrow field of view interactive scenarios, triangular pyramids can be used to further reduce computing power overhead, while for wide field of view panoramic scenarios, pentagonal pyramids and hexagonal pyramids can be used to improve the uniformity of lateral coverage.

[0158] All types of pyramids are equivalent implementations of this invention, with the pyramid axis corresponding to the user's walking and line of sight, and the base angle matching the field of view. Under the constraint that the volume is always equal to the preloaded space reference volume, the pyramid height and base area are dynamically adjusted.

[0159] In summary, the method in the above embodiments sets a fixed preload space baseline volume based on the computing power and memory parameters of the XR device, performs voxel segmentation of the virtual scene and marks the memory occupation of each voxel, constructs a tilted quadrangular pyramid along the user's current orientation as the preload enclosing space, and dynamically adjusts the pyramid height and base area by combining real-time collected user position, walking speed and orientation change rate to always maintain a constant total volume of preload space to adapt to the device's memory carrying capacity limit, introduces a user movement turnaround rate weighted matching adjustment dimension, and reduces the computational overhead caused by frequent form switching with a preset hysteresis threshold, and adopts an incremental loading and unloading mechanism to retain the state of overlapping resources. It can match the user's movement trend to schedule virtual resources under limited memory conditions, reduce resource loading response latency, reduce the memory occupation of invalid preload, and improve the smoothness of virtual scene operation and the rationality and hit rate of resource pre-scheduling loading.

[0160] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for preloading XR virtual resources, characterized in that, include: Based on the computing power and memory parameters of the target XR device, a fixed preload space baseline volume is set and stored in the XR device's local configuration module; Real-time data collection of the user's current spatial position, instantaneous travel speed, and instantaneous orientation change rate within the virtual scene to obtain real-time motion pose parameters; Construct an initial quadrangular pyramid with a tilting posture along the user's current orientation, determine the triangular cross-section of the pyramid containing the pyramid axis, and ensure that the cross-section does not coincide with any of the lateral edges of the initial quadrangular pyramid, so that the user's current spatial position falls on the central axis of the cross-section; Using the preloaded space reference volume as a constraint, the height of the quadrangular pyramid is adjusted based on the instantaneous travel speed, and the base area of ​​the quadrangular pyramid is adjusted accordingly based on the instantaneous orientation change rate, and the volume of the quadrangular pyramid is always equal to the preloaded space reference volume. Traverse all resource units in the virtual scene, filter all resource units that fall within the space enclosed by the adjusted quadrangular pyramid, and generate a preloaded resource list. The corresponding virtual resources are loaded into the device memory according to the preloaded resource list, and the quadrangular pyramid enclosed space and the preloaded resource list are synchronously iterated as the user pose data is updated in real time.

2. The XR virtual resource preloading method according to claim 1, characterized in that, During the preloading phase, a fixed preload space is set to remember the volume. Simultaneously, the virtual dimensions of the virtual scene are acquired. Let the length, width, and height be x, y, and z, respectively. The voxel size a³ is set synchronously, ensuring that x, y, and z are all divisible by a. The virtual scene is then divided using the voxel size. For each voxel, obtain the memory required for loading each voxel, and mark the memory required for loading each voxel on the corresponding voxel. After setting a fixed preload space reference volume, the initial square pyramid shape and the number of voxels are determined. The initial square pyramid is determined to be a pyramid structure of approximately square pyramid shape composed of several voxels. ； In the formula: n represents the number of voxel stacking layers in the initial square pyramid; V represents the set preload space reference volume; When the above equation holds true, the initial square pyramid contains the following number of voxels: If the length, width, and height of the initial quadrangular pyramid are all n×a, and the vertex of the initial quadrangular pyramid is located on the perpendicular line from the center point of the base of the initial quadrangular pyramid.

3. The XR virtual resource preloading method according to claim 2, characterized in that, The preloaded space reference volume V is determined through the following steps: Based on the memory required for loading each voxel, sort the voxels to obtain the voxel queue. Extract the first 1 / 4 to 1 / 5 of the voxel queue, simultaneously identify the position of each extracted voxel in the virtual scene, and construct the minimum bounding box of all extracted voxels based on the identified positions, using the volume of this minimum bounding box as V.

4. The XR virtual resource preloading method according to claim 1, characterized in that, The frequency of collecting the user's current spatial position, instantaneous travel speed, and instantaneous orientation change rate in the virtual scene is consistent with the frequency of acquiring motion pose parameters. The initial frequency is set to 0.1 s / time, and is dynamically adjusted within the range of ±0.05 s / time according to the instantaneous travel speed.

5. The XR virtual resource preloading method according to claim 1, characterized in that, After the initial square pyramid is constructed, it is adjusted to a tilted position along the user's current orientation: The real-time motion pose parameters are based on the user's movement path established by continuously acquiring the user's position coordinates in the virtual scene, and the user movement path contains continuously acquired user position coordinate nodes. Get the path segment formed by the latest two nodes on the user's movement path. and based on Determine the direction of movement; Tilting the initial square pyramid so that the axis passing through the vertex of the initial square pyramid is aligned with... Overlapping, and based on The determined direction of movement is directly opposite the center point of the base of the initial square pyramid after it has tipped over; Based on the intersection of the base and vertex of a square pyramid, the axis passing through the vertex is denoted as... The captured data will be determined based on the direction of movement. The two ends are set as the start and end points. The initial position of the user in the quadrangular pyramid in the virtual scene is set to 3 / 4 of the distance from the end point and 1 / 4 of the distance from the start point, and is dynamically adjusted based on the user's current turnaround rate in the virtual scene.

6. The XR virtual resource preloading method according to claim 1, characterized in that, When adjusting the height of the pyramid based on the instantaneous travel speed, the base area of ​​the pyramid is determined by combining the constraint that the volume of the pyramid is always equal to the preloaded space reference volume. The adjustment of the pyramid height based on instantaneous travel speed follows: ； In the formula: The forward extension height of the square pyramid; Instantaneous speed; This refers to the average loading response latency per resource unit of an XR device. The maximum travel acceleration set for users in a virtual environment; This is the adjusted height of the square pyramid; This is the initial reference height of the square pyramid; This is the adjusted base area of ​​the square pyramid; Preload space reference volume; Among them, the faster the instantaneous speed, the greater the forward extension height of the square pyramid, the higher the total height, and the smaller the corresponding base area.

7. The XR virtual resource preloading method according to claim 1, characterized in that, When adjusting the base area of ​​the pyramid based on the instantaneous orientation change rate, the height of the pyramid is determined by combining the constraint that the volume of the pyramid is always equal to the preloaded space reference volume: The adjustment of the base area of ​​the quadrangular pyramid based on the instantaneous orientation change rate follows the following: ； In the formula: The lateral expansion angle of the base of the square pyramid; The instantaneous orientation change rate; This refers to the average loading response latency per resource unit of an XR device. The maximum user viewpoint acceleration set for the virtual scene; This is the adjusted base area of ​​the square pyramid; The base area of ​​the initial square pyramid; The horizontal field of view angle is the reference angle for XR equipment; This is the adjusted height of the square pyramid; Preload space reference volume; Among them, the faster the instantaneous orientation change rate, the larger the lateral expansion angle of the base of the square pyramid, the larger the total base area, and the smaller the corresponding height.

8. A method for preloading XR virtual resources according to claim 6 or 7, characterized in that, During the dynamic adjustment process, the square pyramid selects either a height adjustment logic based on instantaneous travel speed or a base area adjustment logic based on instantaneous orientation change rate to perform a shape update. The selection logic is as follows: The target cone height is calculated by combining the current instantaneous travel speed. The absolute value of the difference between the current pyramid height and the target cone height is calculated, and the ratio of this difference to the current pyramid height is used to obtain the height deviation rate. The target base area is calculated by combining the current instantaneous orientation change rate. The absolute value of the difference between the current base area of ​​the pyramid and the target base area is calculated, and the ratio of this difference to the current base area of ​​the pyramid is used to obtain the base area deviation rate. Obtain the current user's motion return rate, use the return rate as the lateral weight coefficient, and use the difference between 1 and the return rate as the axial weight coefficient. Weight the bottom area deviation rate and the height deviation rate respectively to obtain the lateral weighted deviation value and the axial weighted deviation value. Compare the values ​​of the two sets of weighted deviation values, and select the dimension corresponding to the larger weighted deviation value for this morphological update; When the absolute value of the difference between the two sets of weighted deviation values ​​is less than the preset hysteresis threshold, the adjustment logic used in the previous frame is maintained as the calculation rule for this morphological update.

9. The XR virtual resource preloading method according to claim 1, characterized in that, When filtering all resource units that fall within the space enclosed by the adjusted quadrangular pyramid, the voxels obtained by voxel segmentation in the virtual scene are used as the smallest unit. All complete voxels contained in the quadrangular pyramid are identified, and the resource units corresponding to each voxel are used as preloading targets to generate a preloaded resource list.

10. The XR virtual resource preloading method according to claim 1, characterized in that, The steps for synchronously iterating the bounded space of the quadrangular pyramid and the preloaded resource list in real time with the user pose data include: After each adjustment of the quadrangular pyramid shape and generation of the adjusted preloaded resource list, the preloaded resource list before adjustment and the preloaded resource list after adjustment are compared. The overlapping resource units in the two lists are retained and their loading status is maintained. All resource units in the pre-loaded resource list that do not belong to the overlapping part will be removed from the loading process; Perform loading operations on newly added resource units that do not belong to the overlapping part in the adjusted preloaded resource list; After all resource units in the preloaded resource list have been loaded after the adjustment, the process will proceed to the next step of determining overlapping resource units after the adjustment of the quadrangular pyramid shape.

Images